The Research And Application On Traffic Identification Method Based On Deep Packet Inspection

As the characteristics of good robustness and high performance, etc, P2 P technology has become one of the develop rapidest network technology in recent years. And it was widely used in file sharing, media streaming and other fields. However, with the P2 P application software increasing hardly, the proportion of P2 P traffic continually growing. This situation not only leads to a decline in quality of Internet service, sharp consumption of network bandwidth resources, but also makes the piracy, privacy disclosure, network security issues become increasingly prominent.In addition, Vo IP also has been developed rapidly with the advantages of good openness, flexibility and low cost, etc. The emergence of Vo IP application software makes the traditional PSTN business suffered serious impact and challenge. Moreover, under the drive of economic interests, illegal Vo IP service providers also caused great interference on telecommunications market.In order to ensure the rights in network business implementation and management of network traffic, how to identify all kinds of traffic quickly and effectively has become a urgent and important matter for Internet service providers.This paper aims at the research on problems of traffic identification and control. Paper analyzed the working principle of P2 P and Vo IP application software and traffic identification and interference technologies, studied the advantages and disadvantages of the commonly used traffic identification method. And on this basis, paper proposed and designed a traffic identification and control system which based on DPI technology. In the process of designing system module, paper studied the AC automata algorithm and put forward an improvement to the algorithm which raised the matching efficiency. The main research work of this paper includes the following aspects:1. Improved the feature matching algorithm of DPI traffic identification technology which is called Aho-Corasicks algorithm. The improved algorithm increased the efficiency by reducing the times of state transition.2. Analysing the working principle of the existed traffic identification technology. Aimed at the situation that inlet flow is separate form business flow and the business flow does not have obvious characteristic signature, proposing a new technique that combined with session affinity and DPI which is called deeply analytical method. This method can identify the specific business flow quickly.3. Combining with the existed technical solutions, paper proposed and designed a traffic identification and control system that based on DPI technology. Paper presented the overall structure of the system, analyzed the realization of the modules, and verified the accuracy and effectiveness of the traffic identification and control system by experiments finally.